Perovskite solar cells are highly efficient. However, there are still issues with long-term stability and upscaling. A KIT study shows how this can be changed.
Artificial intelligence supports the monitoring and optimization of perovskite solar cell production.
Foto: Markus Breig, KIT / Grafik: Felix Laufer, KIT
The Karlsruhe Institute of Technology (KIT) aims to make perovskite solar cells commercially viable. In a study published in the special interest magazine Energy and Environmental Science, KIT Professor Ulrich Wilhelm Paetzold and team show that deep learning – a machine learning method that uses neural networks – is a crucial tool for improving the measurement technology required for the commercial production of perovskite solar cells.
“Based on measurement data collected during production, machine learning can be used to identify process errors before the solar cells are completed,” says Felix Laufer, a research associate at the KIT's Light Technology Institute and the study's first author. He adds that the speed and efficiency of the method significantly improves data analysis. Additional research methods are not necessary.
A Step Towards Industrial Applicability
Perovskite solar cells already demonstrate high efficiencies in the conversion of solar energy into electrical energy and can be produced cost-effectively. In addition, the cells can be made thin and flexible. However, according to Paetzold, challenges remain in terms of long-term stability and upscaling to large areas.
“We are showing how process fluctuations can be quantitatively analyzed by expanding the characterization methods with machine learning techniques,” says the physicist. This ensures high material quality and layer homogeneity over large areas and many batches. And that is a ‘decisive step towards industrial applicability’.
This is a partly automated translation of this german article.